7
$\begingroup$

Let $(M^n,g)$ be a compact Riemannian manifold with boundary $\partial M=N$. Suppose $|Rm_g| \le C_1$ on $M$ and the second fundamental form of $N$ is bounded by $C_2$. Moreover, there exists a constant $\epsilon$ such that $\exp_x(t n_x) \in M\backslash N$ for any $t \in (0,\epsilon]$ and $x \in N$, where $n_x$ is the inner unit normal vector.

Can we find a constant $L=L(C_1,C_2,\epsilon)>0$ such that the map $$ F(x,t)=\exp_x(t n_x) $$ on $N \times [0,L)$ is an embedding?

$\endgroup$
1
  • $\begingroup$ I recommend looking at the classic paper by Heintze and Karcher. $\endgroup$
    – Deane Yang
    Commented Dec 5, 2022 at 3:58

1 Answer 1

5
$\begingroup$

The answer is yes.

The map $F$ is defined in $N \times [0,\varepsilon]$; plus it is smooth. Further, your condition on second fundamental form and curvature imply that it is regular in $N \times [0,10\cdot L]$, where $L=L(C_1,C_2,\varepsilon)>0$.

Suppose $F$ is not injective in $N \times [0,L]$. Let $\ell$ be the smallest value such that $F$ is not injective in $N \times [0,\ell]$. In this case there are two points $x,y\in N$ such that $z=\exp_x(\ell \cdot n_x)=\exp_y(\ell\cdot n_y)$. Since $\ell$ is minimal, we have that geodesics $[zx]$ and $[zy]$ point in the opposite directions. It follows that $y=\exp_x(2\cdot \ell\cdot n_x)$ and therefore $\exp_x(10\cdot L n_x)$ is undefined --- a contradiction.

enter image description here

$\endgroup$

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .